"Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Keith Pieper, director of product management at SpotXchange.
Fraud is a big issue. Quality is a bigger issue. Sadly, both are subjective.
There are a growing number of solutions on the market to help us detect fraud and reduce bad inventory. But what we really need is greater transparency, collaboration and less emphasis on proprietary algorithms.
Without these things, ad fraud will continue to grow.
I’ve seen different solutions report wildly different rates of fraudulent impressions for the same inventory – anywhere from 1% to 23%. These large discrepancies make it nearly impossible for suppliers to comply 100% with the “quality” needs of a buyer.
If the industry wants to eliminate fraudulent traffic, buyers and sellers need to agree on what is “bad” traffic, disclosing their unique and specific definitions of “fraudulent traffic” to one another. Transparency and collaboration is required if the industry is going to succeed.
Another problem is the blind chainsaw approach, such as when an advertiser using one solution asks a supplier to remove impressions sourced from “IP addresses for ISPs and data centers.” This request is valid, but simply too broad. If the publisher filters out IP addresses associated with ISPs, it will eliminate the majority of volume from the campaign. Data centers typically host multiple websites, so blocking a single data center would presumably block many websites. And how does one define the IP address of a data center?
This situation could be resolved if the advertiser provided the supplier with a specific list of IP addresses it wants to block. But they won’t. And why not?
Naturally, everyone is trying to protect themselves and maintain a market advantage. While differing methodologies can result in a competitive advantage, they can also result in finger pointing and dissatisfaction.
Who is to blame for bad traffic? Whose methodology should one believe? Is one vendor right and the other wrong?
Taking a blanket approach to the issue leaves you wondering if you are covered, while taking a narrow, calculated approach makes it easier to pinpoint blame. As a result, the conflict around fraud and bots between buyer and seller is growing, not being minimized.
Today’s Biggest Challenge: Bots
Of course, bots, or nonhuman traffic, don’t parade around the Internet with name tags; they must remain hidden for the scam to be effective. Once a bot has been detected, game over, there is no more money to be made, until they start up as a different bot. While bots exhibit unique characteristics and behaviors, they have become tougher to distinguish from real humans. This difficulty has resulted in vendors creating unique and different ways to detect bot traffic, turning what was intended to be a competitive advantage into a disparate and wide range of “accurate” approaches.
A Growing Toolbox
As the problem balloons in size and complexity, so do the tools used to combat the issue. Today companies have DoubleVerify, WhiteOps, Integral Ad Science, Telemetry, Distil, Iovation, Zvelo, Adtricity, Forensiq, Pixalate and Mdot Labs to choose from, just to name a few. While some are designed to protect buyers from fraudulent traffic, others are designed to filter bad inventory from a publisher’s site and others work across the whole ecosystem. Added to the mix are the supply-side platforms and demand-side platforms that are developing their own secret sauce to detect or filter bots and ad fraud.
While the technology to identify fraudulent traffic varies wildly, there are some basic standards to ground the industry and point us in the right direction.
For starters, obscured URLs have a proposed standard on the table – Safe Frame – but it’s logistically complex to implement and is of no value unless a large number of publishers participate. The industry needs to rally around this standard.
Viewability also has the stamp of approval from the Media Rating Council in both display and video, though there are still numerous unaccredited viewability solutions.
Finally, the IAB recently announced Anti-Fraud Principles, a big, broad step in getting everyone on the same page so we can all start speaking the same language.
The Bottom Line
The only way we’re going to combat ad fraud is through industrywide collaboration. This means sharing methodologies and secrets, not stealing them. For example, it is possible for a seller to integrate a buyer’s ad fraud algorithm exclusively for that buyer, yielding only the exact impression a buyer wants. Likewise, sharing a list of known bot IP addresses between buyers and sellers would allow both parties to tackle the issue together.
We’ll probably never come to a universal agreement on what constitutes a bot or bad traffic, but by taking a collaborative approach, we can become more effective at reducing fraud and increasing marketplace efficiencies.